A high degree of self-disclosure in Online Social Networks (OSNs) is associated with several risks. This raises an important question: Why don´t many users protect their personal data more eagerly? We propose that a lack of memory for what information has been disclosed to which audience contributes to this privacy-neglecting behavior in OSNs. We transferred the paradigm of target monitoring to a fictitious OSN and varied the degree of risk associated with self-disclosure. In a 2x2 experiment we varied both audience size (large vs. small) and information intimacy (personal vs. non-personal). We used recognition tests for the association of audience and disclosed information to assess memory performance. Results show that item memory (the memory for what information has been disclosed) exceeded target memory and that target memory improved in vulnerable situations (for large audiences and personal information). Our findings widen the realm of offline memory research and expand our knowledge about which cognitive factors impact privacy-related behavior in online environments.